Showing papers on "Sequence learning published in 2005"

A generalized growing and pruning RBF (GGAP-RBF) neural network for function approximation

Abstract: This work presents a new sequential learning algorithm for radial basis function (RBF) networks referred to as generalized growing and pruning algorithm for RBF (GGAP-RBF). The paper first introduces the concept of significance for the hidden neurons and then uses it in the learning algorithm to realize parsimonious networks. The growing and pruning strategy of GGAP-RBF is based on linking the required learning accuracy with the significance of the nearest or intentionally added new neuron. Significance of a neuron is a measure of the average information content of that neuron. The GGAP-RBF algorithm can be used for any arbitrary sampling density for training samples and is derived from a rigorous statistical point of view. Simulation results for bench mark problems in the function approximation area show that the GGAP-RBF outperforms several other sequential learning algorithms in terms of learning speed, network size and generalization performance regardless of the sampling density function of the training data.

The interaction of the explicit and the implicit in skill learning: a dual-process approach.

Abstract: This article explicates the interaction between implicit and explicit processes in skill learning, in contrast to the tendency of researchers to study each type in isolation. It highlights various effects of the interaction on learning (including synergy effects). The authors argue for an integrated model of skill learning that takes into account both implicit and explicit processes. Moreover, they argue for a bottom-up approach (first learning implicit knowledge and then explicit knowledge) in the integrated model. A variety of qualitative data can be accounted for by the approach. A computational model, CLARION, is then used to simulate a range of quantitative data. The results demonstrate the plausibility of the model, which provides a new perspective on skill learning.

Temporal Sequence Learning, Prediction, and Control: A Review of Different Models and Their Relation to Biological Mechanisms

Abstract: In this review, we compare methods for temporal sequence learning (TSL) across the disciplines machine-control, classical conditioning, neuronal models for TSL as well as spike-timing-dependent plasticity (STDP). This review introduces the most influential models and focuses on two questions: To what degree are reward-based (e.g., TD learning) and correlationbased (Hebbian) learning related? and How do the different models correspond to possibly underlying biological mechanisms of synaptic plasticity? We first compare the different models in an open-loop condition, where behavioral feedback does not alter the learning. Here we observe that reward-based and correlation-based learning are indeed very similar. Machine control is then used to introduce the problem of closed-loop control (e.g., actor-critic architectures). Here the problem of evaluative (rewards) versus nonevaluative (correlations) feedback from the environment will be discussed, showing that both learning approaches are fundamentally different in the closed-loop condition. In trying to answer the second question, we compare neuronal versions of the different learning architectures to the anatomy of the involved brain structures (basal-ganglia, thalamus, and cortex) and the molecular biophysics of glutamatergic and dopaminergic synapses. Finally, we discuss the different algorithms used to model STDP and compare them to reward-based learning rules. Certain similarities are found in spite of the strongly different timescales. Here we focus on the biophysics of the different calciumrelease mechanisms known to be involved in STDP.

Adaptive Learning Resources Sequencing in Educational Hypermedia Systems

Abstract: Adaptive learning resources selection and sequencing is recognized as among the most interesting research questions in adaptive educational hypermedia systems (AEHS). In order to adaptively select and sequence learning resources in AEHS, the definition of adaptation rules contained in the Adaptation Model, is required. Although, some efforts have been reported in literature aiming to support the Adaptation Model design by providing AEHS designers direct guidance or semi-automatic mechanisms for making the design process less demanding, still it requires significant effort to overcome the problems of inconsistency, confluence and insufficiency, introduced by the use of rules. Due to the problems of inconsistency and insufficiency of the defined rule sets in the Adaptation Model, conceptual “holes” can be generated in the produced learning resource sequences (or learning paths). In this paper, we address the design problem of the Adaptation Model in AEHS proposing an alternative sequencing method that, instead of generating the learning path by populating a concept sequence with available learning resources based on pre-defined adaptation rules, it first generates all possible learning paths that match the learning goal in hand, and then, adaptively selects the desired one, based on the use of a decision model that estimates the suitability of learning resources for a targeted learner. In our simulations we compare the learning paths generated by the proposed methodology with ideal ones produced by a simulated perfect rule-based AEHS. The simulation results provide evidence that the proposed methodology can generate almost accurate learning paths avoiding the need for defining complex rule sets in the Adaptation Model of AEHS.

On-Line Sequential Extreme Learning Machine

Abstract: The primitive Extreme Learning Machine (ELM) [1, 2, 3] with additive neurons and RBF kernels was implemented in batch mode. In this paper, its sequential modification based on recursive least-squares (RLS) algorithm, which referred as Online Sequential Extreme Learning Machine (OS-ELM), is introduced. Based on OS-ELM, Online Sequential Fuzzy Extreme Learning Machine (Fuzzy-ELM) is also introduced to implement zero order TSK model and first order TSK model. The performance of OS-ELM and Fuzzy-ELM are evaluated and compared with other popular sequential learning algorithms, and experimental results on some real benchmark regression problems show that the proposedOnlineSequentialExtreme Learning Machine (OS-ELM) produces better generalization performance at very fast learning speed.

Do children with developmental dyslexia have an implicit learning deficit

Abstract: Objective: The purpose of this study was to investigate the effects of specific types of tasks on the efficiency of implicit procedural learning in the presence of developmental dyslexia (DD). Methods: Sixteen children with DD (mean (SD) age 11.6 (1.4) years) and 16 matched normal reader controls (mean age 11.4 (1.9) years) were administered two tests (the Serial Reaction Time test and the Mirror Drawing test) in which implicit knowledge was gradually acquired across multiple trials. Although both tests analyse implicit learning abilities, they tap different competencies. The Serial Reaction Time test requires the development of sequential learning and little (if any) procedural learning, whereas the Mirror Drawing test involves fast and repetitive processing of visuospatial stimuli but no acquisition of sequences. Results: The children with DD were impaired on both implicit learning tasks, suggesting that the learning deficit observed in dyslexia does not depend on the material to be learned (with or without motor sequence of response action) but on the implicit nature of the learning that characterises the tasks. Conclusion: Individuals with DD have impaired implicit procedural learning.

Individual differences in working memory capacity and learning: evidence from the serial reaction time task.

Abstract: High and low working memory (WM) capacity individuals performed the serial reaction time task under both incidental and intentional learning conditions to determine the role of WM capacity in the learning of sequential information. WM capacity differences emerged in conditions of intentional but not incidental learning, indicating that individual differences in WM capacity occur in tasks requiring some form of control, with little difference appearing on tasks that required relatively automatic processing. Furthermore, an index of learning was significantly related to a measure of general fluid intelligence under intentional conditions only. Thus, the degree of learning was significantly related to higher order cognition, but only when intentional processing was emphasized.

The neural correlates of implicit and explicit sequence learning: Interacting networks revealed by the process dissociation procedure.

Abstract: In two H2(15)O PET scan experiments, we investigated the cerebral correlates of explicit and implicit knowledge in a serial reaction time (SRT) task. To do so, we used a novel application of the Process Dissociation Procedure, a behavioral paradigm that makes it possible to separately assess conscious and unconscious contributions to performance during a subsequent sequence generation task. To manipulate the extent to which the repeating sequential pattern was learned explicitly, we varied the pace of the choice reaction time task-a variable that is known to have differential effects on the extent to which sensitivity to sequence structure involves implicit or explicit knowledge. Results showed that activity in the striatum subtends the implicit component of performance during recollection of a learned sequence, whereas the anterior cingulate/mesial prefrontal cortex (ACC/MPFC) supports the explicit component. Most importantly, we found that the ACC/MPFC exerts control on the activity of the striatum during retrieval of the sequence after explicit learning, whereas the activity of these regions is uncoupled when learning had been essentially implicit. These data suggest that implicit learning processes can be successfully controlled by conscious knowledge when learning is essentially explicit. They also supply further evidence for a partial dissociation between the neural substrates supporting conscious and nonconscious components of performance during recollection of a learned sequence.

Children's learning.

Abstract: A new field of children's learning is emerging. This new field differs from the old in recognizing that children's learning includes active as well as passive mechanisms and qualitative as well as quantitative changes. Children's learning involves substantial variability of representations and strategies within individual children as well as across different children. The path of learning involves the introduction of new approaches as well as changes in the frequency of prior ones. The rate and the breadth of learning tend to occur at a human scale, intermediate between the extremes depicted by symbolic and connectionist models. Learning has many sources; one that is particularly promising for educational purposes is self-explanations. Overall, contemporary analyses show that learning and development have a great deal in common.

On the Benefits of not Trying: Brain Activity and Connectivity Reflecting the Interactions of Explicit and Implicit Sequence Learning

Abstract: Under certain circumstances, implicit, automatic learning may be attenuated by explicit memory processes. We explored the brain basis of this phenomenon in a functional magnetic resonance imaging (fMRI) study of motor sequence learning. Using a factorial design that crossed subjective intention to learn (explicit versus implicit) with sequence difficulty (a standard versus a more complex alternating sequence), we show that explicit attempts to learn the difficult sequence produce a failure of implicit learning and, in a follow-up behavioural experiment, that this failure represents a suppression of learning itself rather than of the expression of learning. This suppression is associated with sustained right frontal activation and attenuation of learning-related changes in the medial temporal lobe and the thalamus. Furthermore, this condition is characterized by a reversal of the fronto-thalamic connectivity observed with unimpaired implicit learning. The findings demonstrate a neural basis for a well-known behavioural effect: the deleterious impact of an explicit search upon implicit learning.

Neural correlates of encoding and expression in implicit sequence learning.

Abstract: In the domain of motor learning it has been difficult to separate the neural substrate of encoding from that of change in performance. Consequently, it has not been clear whether motor effector areas participate in learning or merely modulate changes in performance. Here, using a variant of the serial reaction time task that dissociated these two factors, we report that encoding during procedural motor learning does engage cortical motor areas and can be characterized by distinct early and late encoding phases. The highest correlation between activation and subsequent changes in motor performance was seen in the motor cortex during early encoding, and in the basal ganglia during the late encoding phase. Our results show that rapid encoding during procedural motor learning involves several distinct processes, and is represented primarily within motor system structures.

Attentional load and implicit sequence learning.

Abstract: A widely employed conceptualization of implicit learning hypothesizes that it makes minimal demands on attentional resources. This conjecture was investigated by comparing learning under single-task and dual-task conditions in the sequential reaction time (SRT) task. Participants learned probabilistic sequences, with dual-task participants additionally having to perform a counting task using stimuli that were targets in the SRT display. Both groups were then tested for sequence knowledge under single-task (Experiments 1 and 2) or dual-task (Experiment 3) conditions. Participants also completed a free generation task (Experiments 2 and 3) under inclusion or exclusion conditions to determine if sequence knowledge was conscious or unconscious in terms of its access to intentional control. The experiments revealed that the secondary task impaired sequence learning and that sequence knowledge was consciously accessible. These findings disconfirm both the notion that implicit learning is able to proceed normally under conditions of divided attention, and that the acquired knowledge is inaccessible to consciousness. A unitary framework for conceptualizing implicit and explicit learning is proposed.

The role of salience in the extraction of algebraic rules.

Abstract: Recent research suggests that humans and other animals have sophisticated abilities to extract both statistical dependencies and rule-based regularities from sequences. Most of this research stresses the flexibility and generality of such processes. Here the authors take up an equally important project, namely, to explore the limits of such processes. As a case study for rule-based generalizations, the authors demonstrate that only repetition-based structures with repetitions at the edges of sequences (e.g., ABCDEFF but not ABCDDEF) can be reliably generalized, although token repetitions can easily be discriminated at both sequence edges and middles. This finding suggests limits on rule-based sequence learning and new interpretations of earlier work alleging rule learning in infants. Rather than implementing a computerlike, formal process that operates over all patterns equally well, rule-based learning may be a highly constrained and piecemeal process driven by perceptual primitives--specialized type operations that are highly sensitive to perceptual factors.

Stacked sequential learning

Abstract: We describe a new sequential learning scheme called "stacked sequential learning". Stacked sequential learning is a meta-learning algorithm, in which an arbitrary base learner is augmented so as to make it aware of the labels of nearby examples. We evaluate the method on several "sequential partitioning problems", which are characterized by long runs of identical labels. We demonstrate that on these problems, sequential stacking consistently improves the performance of nonsequential base learners; that sequential stacking often improves performance of learners (such as CRFs) that are designed specifically for sequential tasks; and that a sequentially stacked maximum-entropy learner generally outperforms CRFs.

A Cognitive Approach to Instructional Design for Multimedia Learning

Abstract: Introduction Cognitive theory is borne from the relatively new interdisciplinary field of cognitive science. Cognitive science studies the nature of the mind by drawing from research in a number of areas including psychology, neuroscience, artificial intelligence, computer science, linguistics, philosophy, and biology. The term cognitive refers to perceiving and knowing, and cognitive scientists seek to understand mental processes such as perceiving, thinking, remembering, understanding language, and learning (Stillings, Weisler, Chase, Feinstein, Garfield, & Rissland, 1995). As such, cognitive science can provide powerful insight into human nature, and, more importantly, the potential of humans to develop increasingly powerful information technologies. This paper addresses the problem that much of what we are currently seeing in multimedia instruction instruction may actually hinder the learning that it claims to promote and then discusses possible ways to improve it. I introduce several well-known assumptions of cognitive science, which provide a framework for applying empirical theories of cognition and learning that improve multimedia instruction and assist humans in learning more effectively. The cognitive theories discussed in the paper include the Theory of Working Memory, Dual Encoding Theory, Cognitive Load Theory, ACT-R Production System Theory, and the Cognitive Theory of Multimedia Learning. Since most instructors have either already been tasked with creating multimedia instruction, or soon will be, this paper is aimed as much at the general practitioner of multimedia instruction as it is the experienced e-learning developer. Popular forms of multimedia instruction, such as online learning and the more inclusive computer-based training (CBT), have created many new possibilities for education. They provide new ways of delivering content, and they often promote learner-centered environments that can motivate students and add variety to learning. In this environment, instructional units are often accompanied by a liberal use of multimedia that is intended to add excitement to the lesson and hold the learner's attention. However, visual and auditory components that are intended to stimulate rather than educate do not always make for sound instructional design in multimedia delivery and can quickly become counter-productive to learning. The human mind is limited in the amount of information that it can process (Miller, 1956). Because computer-based training can quickly overwhelm these limited capacities (Sweller, 1988, 1994), it becomes important for the instructional designer to understand the principles of cognitive science and how they apply to effective instructional design for online learning. Concepts, such as working memory, cognitive load, production system theories of knowledge and learning, self-explaining behaviors, and transfer, all become important considerations for the instructional designer who must learn to use technology effectively and intelligently, rather than simply because it is available and seems flashy or exciting. This is especially relevant as education begins to turn to gaming as the latest innovative technology that some educators claim will revolutionize learning. Proponents of gaming in education, however, should remember that similar predictions were made for mimeograph machines, overhead projectors, movies, radios, television, and the computer, only to produce disappointing results after considerable expenditures of money (Cuban, 1996, 2001). One concern should be that using video games as an educational medium may actually decrease learning in comparison to simply presenting the information in a straightforward manner using text and pictures. Until recently, much of what we have seen in multimedia instructional design appears to be based more on intuition than empirically-based research. For example, it might seem that an online activity that uses flashy multimedia and game-like strategies to hold a learner's attention is good. …

Sequence learning: response structure and effector transfer.

Abstract: Two experiments are reported that investigate the response structure and effector transfer of repeated movement sequences. Participants moved a lever to targets sequentially presented on the computer monitor. In Experiment 1 the learning of 10- and 16-element sequences (identical movement pattern) was contrasted. After 1 day of practice the 10-element sequence was organized into fewer subsequences and, thus, performed more rapidly than the 16-element sequence. The imposed organization appeared to be coded in a relatively abstract way, as evidenced by effector transfer that was as good as that on the retention test. In Experiment 2 the 16-element sequence was studied after more extensive practice. By the end of 4 days of practice the participants produced relatively seamless responses void of obvious transitions between subsequences, but the control of the movement was less effector independent than observed earlier in practice. The results suggest that the process of consolidating the sequence, which led ...

Implicit learning of nonlocal musical rules: implicitly learning more than chunks.

Abstract: Dominant theories of implicit learning assume that implicit learning merely involves the learning of chunks of adjacent elements in a sequence. In the experiments presented here, participants implicitly learned a nonlocal rule, thus suggesting that implicit learning can go beyond the learning of chunks. Participants were exposed to a set of musical tunes that were all generated using a diatonic inversion. In the subsequent test phase, participants either classified test tunes as obeying a rule (direct test) or rated their liking for the tunes (indirect test). Both the direct and indirect tests were sensitive to knowledge of chunks. However, only the indirect test was sensitive to knowledge of the inversion rule. Furthermore, the indirect test was overall significantly more sensitive than the direct test, thus suggesting that knowledge of the inversion rule was below an objective threshold of awareness.

Towards an explicit account of implicit learning.

Abstract: Purpose of reviewThe human brain supports acquisition mechanisms that can extract structural regularities implicitly from experience without the induction of an explicit model. Reber defined the process by which an individual comes to respond appropriately to the statistical structure of the input e

Effector dependent sequence learning in the serial RT task

Abstract: At least five earlier studies could not find effector-dependent learning in the keying version of the serial reaction time (RT) task. Experiment 1 examined whether effector-dependent learning occurs when participants practice the serial RT task with three fingers of one hand for about 1,300 sequence repetitions instead of the more common 50–100 repetitions. The results confirm that, following extended practice, sequence learning produces an effector-dependent component. Specifically, an unpracticed hand executed a practiced sequence slower than a practiced hand. However, Experiment 2 showed that effector-dependent sequence learning develops only when fingers of one hand are used, suggesting that effector-dependent sequence learning involves adjustment to the mechanical interactions between the fingers of one hand. In addition, when sequences had been practiced with one hand, mirror versions of the practiced sequences in both experiments showed moderate transfer. But when practiced with two hands no transfer to a mirrored version of the sequence was observed. This suggests that only practice with one hand produces a representation that facilitates the execution of mirror sequences. Generally, the same results were found in more or less aware participants, congruent with the idea that the effector-dependent representation and the representation allowing transfer to mirror sequences are implicit.

Evolino: hybrid neuroevolution / optimal linear search for sequence learning

Abstract: Current Neural Network learning algorithms are limited in their ability to model non-linear dynamical systems. Most supervised gradient-based recurrent neural networks (RNNs) suffer from a vanishing error signal that prevents learning from inputs far in the past. Those that do not, still have problems when there are numerous local minima. We introduce a general framework for sequence learning, EVOlution of recurrent systems with LINear outputs (Evolino). Evolino uses evolution to discover good RNN hidden node weights, while using methods such as linear regression or quadratic programming to compute optimal linear mappings from hidden state to output. Using the Long Short-Term Memory RNN Architecture, the method is tested in three very different problem domains: 1) context-sensitive languages, 2) multiple superimposed sine waves, and 3) the Mackey-Glass system. Evolino performs exceptionally well across all tasks, where other methods show notable deficiencies in some.

Sequence learning under dual-task conditions: alternatives to a resource-based account.

Abstract: In two experiments with the serial reaction-time task, participants were presented with deterministic or probabilistic sequences under single- or dual-task conditions. Experiment 1 showed that learning of a probabilistic structure was not impaired over a first session by performing a counting task, but that such an interference arose over a second session, when the knowledge was tested under single-task conditions. In contrast, the effects of the secondary task arose earlier for participants exposed to deterministic sequences. This difference between deterministic and probabilistic sequences disappeared in Experiment 2, where the counting task was performed on tones associated to the locations. Comparisons between sessions indicated that the secondary task affected not only the expression but also the acquisition of sequence learning, and that greater interference was observed in those conditions that yielded more explicit knowledge. These results suggest that the effects of a dual task on the measures of implicit sequence learning may be partly due to the intrusion of explicit knowledge and partly due to the disruption of the sequence produced by the inclusion of random events.

Impaired implicit sequence learning in depression: a probe for frontostriatal dysfunction?

Abstract: Background. Implicit learning through motor sequencing tasks is sensitive to basal ganglia dysfunction. Consequently, it is ideally suited for testing elements of the frontostriatal model of major depression and performance can be related to key clinical, neuropsychological, vascular and biochemical data. Method. Twenty-one subjects with moderate to severe unipolar depression and 21 age-, sex- and education-matched controls were recruited. Clinical, vascular and biochemical data were recorded. Subjects were administered a battery of neuropsychological tests that assessed speed of processing, working memory, learning, memory, language, perceptual organization and executive functioning. Additionally, subjects were administered a motor sequencing implicit learning task. Implicit learning is assumed when reaction times improve during the sequenced condition as compared to the pseudo-random baseline condition. Results. The rate of implicit learning in persons with depression was only half that of control subjects (3·6% v. 7·3%). Lower rates of implicit learning in patients were associated with poorer performance on neuropsychological tests of visuomotor speed and mental flexibility, longer duration of depressive episode and severity of acute stress. In a small number of subjects, poorer performance was also related to past suicide attempt. Conclusions. Impaired implicit learning in persons with depression is consistent with frontostriatal dysfunction. Performance is related to some clinical characteristics and to neuropsychological functioning on tests of visuomotor speed and mental flexibility.

A Hierarchy of Associations in Hippocampo-Cortical Systems: Cognitive Maps and Navigation Strategies

Abstract: In this letter we describe a hippocampo-cortical model of spatial processing and navigation based on a cascade of increasingly complex associative processes that are also relevant for other hippocampal functions such as episodic memory. Associative learning of different types and the related pattern encoding-recognition take place at three successive levels: (1) an object location level, which computes the landmarks from merged multimodal sensory inputs in the parahippocampal cortices; (2) a subject location level, which computes place fields by combination of local views and movement-related information in the entorhinal cortex; and (3) a spatiotemporal level, which computes place transitions from contiguous place fields in the CA3-CA1 region, which form building blocks for learning temporospatial sequences.At the cell population level, superficial entorhinal place cells encode spatial, context-independent maps as landscapes of activity; populations of transition cells in the CA3-CA1 region encode context-dependent maps as sequences of transitions, which form graphs in prefrontal-parietal cortices. The model was tested on a robot moving in a real environment; these tests produced results that could help to interpret biological data. Two different goal-oriented navigation strategies were displayed depending on the type of map used by the system.Thanks to its multilevel, multimodal integration and behavioral implementation, the model suggests functional interpretations for largely unaccounted structural differences between hippocampo-cortical systems. Further, spatiotemporal information, a common denominator shared by several brain structures, could serve as a cognitive processing frame and a functional link, for example, during spatial navigation and episodic memory, as suggested by the applications of the model to other domains, temporal sequence learning and imitation in particular.

Measures of working memory, sequence learning, and speech recognition in the elderly.

Abstract: This study describes the measurement of 2 cognitive functions, working-memory capacity and sequence learning, in 2 groups of listeners: young adults with normal hearing and elderly adults with impa...